You know, when I did a search on "teoma search engine technology" on several search engines (including google and teoma of course), the only one that came up with a decent result was Lookle, with this result from about.com: Review of Teoma Search Engine - Tasty New Search Engines Part 2. Now Lookle is somewhat innovative, and copycats google to no end, and were apparently threatened by google (trademark, not patent) last June. As far as I know, nothing came of it because google decided not to pursue.
Now as I see it there's two kinds of companies out there: those that are too busy innovating to sue, and those too busy suing to innovate. Adobe seems to have joined the Dark side; But Google's been making all sorts of cool changes lately, enough to make you think they hired more linux geeks, not more lawyers. So I doubt anything like a patent suit will cross their minds. When it does, I switch search engines. It hasn't for, what, seven years?
And THAT's why google's *my* favorite search engine.
I'm working on a project to actually do this well. I'm not claiming it's all it can be, but it's better than a lot of the nat. lang. support bits out there.
Can I ask you guys to bang on it for a while if you use WebLogic or Tuxedo or other BEA products? It's at the
BEA's support site. It clearly won't answer things like "I just got a computer, what do I do?" but it's not aimed at that. It's supposed to help sysadmins and knowledgeable users like many slashdot readers get to their info quicker. Give it a whirl, eh?
For code similarity (Computer Science applications), there is an excellent piece of software from UC Berkeley called MOSS. For the kind of text similarity this guy was doing, you don't have to home-grow your software either. Just look at findsame. Both use some rather sophisticated algorithms.
As with security through obscurity, cheating detectors need to do only well enough that fooling them is more work than doing your own work.
A six-word match doesn't meet that criterion. I'm not surprised that he didn't find many matches this year, but I'd be surprised if some students didn't get around the check by, say, changing tense throughout the passage.
Preventing cheating should of course be easier than prosecuting it. Smaller class size, with instructors teaching somewhat different --more up to date-- material is more expensive, but better for the students; I have a feeling that if the instructor thus knows her students and is percieved to care whether they learn, that they will be less likely to cheat.
Good job to the students for teaching their prof a hard lesson -- he needs to care and be involved. Kudos to the prof for caring and going after them.
Slashdot Mirror
You know, when I did a search on "teoma search engine technology" on several search engines (including google and teoma of course), the only one that came up with a decent result was Lookle, with this result from about.com: Review of Teoma Search Engine - Tasty New Search Engines Part 2. Now Lookle is somewhat innovative, and copycats google to no end, and were apparently threatened by google (trademark, not patent) last June. As far as I know, nothing came of it because google decided not to pursue.
Now as I see it there's two kinds of companies out there: those that are too busy innovating to sue, and those too busy suing to innovate. Adobe seems to have joined the Dark side; But Google's been making all sorts of cool changes lately, enough to make you think they hired more linux geeks, not more lawyers. So I doubt anything like a patent suit will cross their minds. When it does, I switch search engines. It hasn't for, what, seven years?
And THAT's why google's *my* favorite search engine.
Gremio
I'm working on a project to actually do this well. I'm not claiming it's all it can be, but it's better than a lot of the nat. lang. support bits out there.
Can I ask you guys to bang on it for a while if you use WebLogic or Tuxedo or other BEA products? It's at the BEA's support site. It clearly won't answer things like "I just got a computer, what do I do?" but it's not aimed at that. It's supposed to help sysadmins and knowledgeable users like many slashdot readers get to their info quicker. Give it a whirl, eh?
Gremio
For code similarity (Computer Science applications), there is an excellent piece of software from UC Berkeley called MOSS. For the kind of text similarity this guy was doing, you don't have to home-grow your software either. Just look at findsame. Both use some rather sophisticated algorithms.
As with security through obscurity, cheating detectors need to do only well enough that fooling them is more work than doing your own work. A six-word match doesn't meet that criterion. I'm not surprised that he didn't find many matches this year, but I'd be surprised if some students didn't get around the check by, say, changing tense throughout the passage.
Preventing cheating should of course be easier than prosecuting it. Smaller class size, with instructors teaching somewhat different --more up to date-- material is more expensive, but better for the students; I have a feeling that if the instructor thus knows her students and is percieved to care whether they learn, that they will be less likely to cheat.
Good job to the students for teaching their prof a hard lesson -- he needs to care and be involved. Kudos to the prof for caring and going after them.
Gremio
--